The Latent Load Arbitrage: Expert Insights on Thermal Storage as a Trading Asset

The Strategic Imperative: Why Latent Load Arbitrage Demands Your Attention Now

For decades, thermal storage was viewed primarily as a demand-side management tool—a way to shave peak loads and reduce demand charges. That perspective is increasingly outdated. In deregulated electricity markets with high renewable penetration, price volatility has become the dominant feature, creating opportunities for assets that can shift consumption across time. Latent load arbitrage—the practice of using thermal storage to buy electricity when prices are low and sell it back (implicitly) by avoiding consumption when prices are high—has matured from a niche engineering tactic into a serious trading strategy. Operators of large commercial buildings, district cooling systems, and industrial refrigeration plants now sit on untapped flexibility that can generate returns comparable to financial instruments, without the counterparty risk of pure speculation.

This article is written for experienced energy professionals who already understand the basics of thermal storage. We will not rehash the physics of latent heat or the types of storage media. Instead, we focus on the arbitrage thesis: how to structure operations, select technologies, and manage risks to capture the spread between low-cost and high-cost periods. We assume familiarity with time-of-use rates, real-time pricing, and ancillary service markets. Our goal is to provide a framework for turning a thermal asset into a profit center, with concrete examples and decision criteria drawn from real projects.

The Scale of the Opportunity

Consider a typical 500,000-square-foot commercial office building with a 1,000-ton chiller plant. If equipped with a thermal storage tank sized for 10,000 ton-hours, the building can shift its entire cooling load off-peak for several hours. In a market like ERCOT, where summer real-time prices can spike from $30/MWh to over $500/MWh, the arbitrage value per megawatt-hour shifted can exceed $200. Over a cooling season, this translates to hundreds of thousands of dollars in gross margin—often exceeding the incremental capital cost of the storage system within two to three years. Yet many operators leave this value on the table because they treat thermal storage as a fixed asset rather than a trading instrument.

We will explore how to change that mindset. From designing control algorithms that optimize against day-ahead and real-time price signals to participating in frequency regulation markets, the path to monetization is clear but requires deliberate engineering and operational discipline. This article provides the roadmap.

Core Frameworks: How Latent Load Arbitrage Works in Practice

At its heart, latent load arbitrage exploits the temporal mismatch between electricity prices and cooling demand. The mechanism is straightforward: during low-price periods (typically overnight or when renewable generation is abundant), the thermal storage system is charged by running chillers or refrigeration equipment to freeze water, ice, or phase-change materials. During high-price periods, the stored cooling capacity is discharged to meet the load, allowing the primary cooling equipment to be turned down or off. The net effect is a reduction in electricity costs equal to the difference between the cost of charging power (at low prices) and the avoided cost of discharging power (at high prices), minus system losses and operational expenses.

Price Spreads and Market Structures

The viability of arbitrage depends on the magnitude and predictability of price spreads. In markets with high renewable penetration, such as California (CAISO) or Germany, intraday price spreads can exceed $100/MWh on a regular basis. These spreads are driven by the duck curve—midday solar overgeneration depressing prices, followed by evening ramp when solar fades and demand peaks. Thermal storage is ideally suited to capture this pattern because cooling loads often correlate with solar availability; the storage can be charged during the midday trough and discharged during the evening peak. However, the exact strategy must be tailored to local market rules, including whether the facility can participate in wholesale markets directly or must rely on retail time-of-use tariffs.

A critical framework is the concept of round-trip efficiency. For a chilled water storage system, round-trip efficiency is typically 90–95% (losses from pump work, heat gain, and temperature stratification). For ice storage, efficiency is lower—around 70–80%—because of the additional energy required to freeze ice at a lower evaporator temperature. Phase-change materials (PCMs) can offer efficiencies in between, depending on the phase-change temperature. The arbitrageur must account for these losses when calculating the effective spread. For example, if the charging price is $30/MWh and the discharging price is $80/MWh, the gross spread is $50/MWh. But with 75% round-trip efficiency, the net spread is only $50 * 0.75 - $30 * 0.25 = $37.5 - $7.5 = $30/MWh, still attractive but significantly lower.

Storage Sizing and Dispatch Strategy

Optimal sizing balances the capital cost of storage against the expected arbitrage revenue. A common heuristic is to size storage to cover the peak cooling load for two to four hours, which captures the highest price spikes without overinvesting. However, with the growth of negative prices (periods where generators pay to export power), there is an emerging opportunity to charge during negative-price events and discharge later, effectively being paid to store energy. This requires a control system that can respond to real-time price signals rather than fixed schedules. Advanced algorithms now incorporate weather forecasts, occupancy patterns, and price predictions to optimize dispatch dynamically. We recommend a model predictive control (MPC) approach that solves a constrained optimization problem over a rolling horizon, typically 24 to 48 hours ahead, balancing thermal comfort constraints against economic objectives.

Execution Workflows: Building a Repeatable Arbitrage Process

Moving from theory to practice requires a structured workflow that integrates market data, system controls, and operational procedures. Based on implementations in several large district cooling plants, we have distilled the process into five stages: (1) market analysis and tariff negotiation, (2) control system design and integration, (3) commissioning and calibration, (4) daily operations and optimization, and (5) performance monitoring and reconciliation. Each stage has specific deliverables and decision points.

Stage 1: Market Analysis and Tariff Selection

Before any hardware is installed, the operator must understand the price environment. For facilities on retail tariffs, the key is to choose a rate structure that rewards load shifting—typically a time-of-use (TOU) rate with significant on-peak/off-peak differentials, or a real-time pricing (RTP) tariff that passes through wholesale prices. In some markets, there are also demand response programs that pay for load reduction during emergency events. The operator should simulate historical price data against the building's load profile to estimate potential savings. We have seen cases where switching from a flat rate to an RTP tariff increased annual arbitrage revenue by 40% without any capital investment. However, RTP also introduces price risk; a hedging strategy (e.g., fixed-price blocks for a portion of consumption) may be prudent.

Stage 2: Control System Design

The control system is the brain of the arbitrage operation. At a minimum, it must be able to receive price signals (either directly from the market or via a third-party data feed), compute an optimal charging/discharging schedule, and communicate setpoints to the chiller plant and storage system. We recommend using an open-standard protocol like BACnet or Modbus for interoperability. The optimization engine should incorporate constraints such as minimum chiller runtime, storage capacity limits, and indoor temperature bounds. Many operators start with a simple rule-based approach (e.g., charge fully during the cheapest 4 hours of the day) and then graduate to MPC. The cost of a basic control upgrade is typically $20,000–$50,000 for a mid-sized plant, with payback periods under one year.

Stage 3: Commissioning and Calibration

Once the control system is installed, a commissioning period of at least two weeks is needed to validate performance. During this phase, the system should be run through a series of test cycles—full charge, full discharge, partial cycles—while measuring energy consumption, storage capacity, and thermal response times. Calibration is critical: sensors for temperature, flow, and power must be accurate to within 1–2% to ensure that the optimization model reflects reality. We have seen projects where a 5% flow measurement error led to a 15% reduction in realized savings. The commissioning team should also verify that the building's thermal comfort is maintained under all discharge scenarios. A common mistake is to assume that the storage can discharge at a constant rate throughout the cycle; in practice, the discharge rate decays as the storage medium warms, requiring the control algorithm to account for this nonlinearity.

Stage 4: Daily Operations

During normal operation, the control system executes the optimized schedule, but human oversight is still essential. An operator should review the day-ahead price forecast each morning, check the storage state of charge, and confirm that the building's load profile matches expectations. In our experience, the biggest operational risk is a sudden change in weather (e.g., an unexpected heat wave) that causes the storage to be depleted too early. The operator should have the authority to override the schedule and activate backup chillers if needed. We recommend establishing clear thresholds: for example, if the indoor temperature rises above 75°F (24°C) and the storage is below 20% state of charge, the system reverts to standard chiller operation. This safety net prevents comfort violations while preserving most of the arbitrage value.

Stage 5: Performance Monitoring

Finally, the operator must track performance metrics to ensure the system is delivering expected returns. Key metrics include: actual versus predicted arbitrage revenue (in dollars), round-trip efficiency (in percent), storage utilization (in percent of capacity used per cycle), and chiller part-load performance. A monthly reconciliation report should compare the facility's electricity costs against a baseline without storage (simulated from the same load and price data). Any deviation greater than 5% should trigger an investigation. We have found that performance tends to degrade over time due to sensor drift, fouling of heat exchangers, or changes in building occupancy. An annual maintenance audit is recommended to recalibrate sensors and clean equipment.

Technology Stack and Economics: Choosing the Right Storage Medium and Control Platform

The choice of thermal storage technology has a profound impact on arbitrage economics. The three main options—chilled water, ice storage, and phase-change materials (PCMs)—each have distinct characteristics in terms of energy density, round-trip efficiency, capital cost, and operational complexity. Understanding these trade-offs is essential for selecting the right system for a given application.

Chilled Water Storage

Chilled water storage is the simplest and most mature technology. It uses a large tank (typically concrete or steel) filled with water that is chilled to around 40°F (4.4°C) during charging. The tank is often stratified, with cold water at the bottom and warm water at the top, separated by a thermocline. Chilled water systems offer high round-trip efficiency (90–95%) and low maintenance, but they have low energy density—about 1–2 kWh per cubic meter. This means they require large physical footprints, which can be a constraint in urban settings. Capital costs range from $50 to $150 per kWh of storage capacity, depending on tank size and site conditions. For large district cooling plants with available space, chilled water is often the most economical choice.

Ice Storage

Ice storage uses the latent heat of fusion of water (334 kJ/kg) to store cooling energy at a much higher density—about 4–5 kWh per cubic meter. This allows for smaller tanks, making ice storage suitable for buildings with limited space. However, the lower evaporator temperature required to freeze ice (typically 20–25°F) reduces chiller efficiency, giving a round-trip efficiency of only 70–80%. Ice storage systems are also more complex, requiring specialized ice-harvesting or coil-freeze equipment. Capital costs are higher, typically $100–$200 per kWh. Despite the efficiency penalty, ice storage can be attractive in markets with very high peak prices, because the higher density allows for longer discharge durations without requiring excessive space.

Phase-Change Materials

PCMs are an emerging category that uses materials with a phase-change temperature tailored to the application—for example, salt hydrates that melt at 45–50°F (7–10°C). This allows the chiller to operate at a higher evaporator temperature, improving efficiency. PCMs offer energy densities between chilled water and ice (2–4 kWh/m³) and round-trip efficiencies of 80–90%. However, the materials themselves can be expensive, and there are concerns about long-term stability and degradation after repeated cycling. Capital costs are currently $150–$300 per kWh, but they are expected to decline as manufacturing scales. PCMs are best suited for retrofit projects where existing chillers cannot operate at ice-making temperatures.

Control Platform Considerations

Regardless of the storage medium, the control platform is critical. We recommend a cloud-based energy management system (EMS) that integrates with the building automation system (BAS) and receives real-time price data via API. Several commercial platforms, such as those from Enel X, CPower, and GridPoint, offer pre-built optimization modules for thermal storage. For large, sophisticated operators, a custom solution using Python or MATLAB with an optimization solver (e.g., Gurobi or CPLEX) may provide better performance. The table below summarizes the key trade-offs.

Scaling the Opportunity: Positioning for Growth and Persistent Revenue

Once a single facility is operating profitably, the natural next step is to scale the strategy across a portfolio of buildings. However, scaling introduces new challenges: coordinating multiple assets, managing diverse market exposures, and maintaining consistent performance. This section addresses how to position a thermal storage arbitrage program for growth, including aggregation strategies, participation in ancillary services, and long-term contracts.

Portfolio Aggregation and Virtual Power Plants

Individual buildings are too small to participate directly in wholesale electricity markets in most jurisdictions. The solution is aggregation—combining multiple thermal storage assets into a virtual power plant (VPP) that can bid into day-ahead and real-time markets, as well as ancillary service markets like frequency regulation and spinning reserves. Several third-party aggregators (e.g., Stem, Tesla, and Autogrid) offer platforms that connect distributed storage assets to wholesale markets, taking a share of the revenue. For operators with a large portfolio (e.g., a university campus or a real estate investment trust), it may be more economical to self-aggregate and hire a market operator. The key is to ensure that the control systems across sites are interoperable and can respond to dispatch signals within seconds. In one case, a portfolio of 12 office buildings with ice storage was aggregated to provide 5 MW of frequency regulation, generating an additional $150,000 per year in revenue beyond arbitrage.

Ancillary Service Markets

Thermal storage can also provide fast-responding capacity for frequency regulation (RegD in PJM, or similar products in other ISOs). Because thermal storage can ramp its charging load up or down quickly (within minutes), it is well-suited for this application. However, the revenue from regulation can be volatile and may conflict with the arbitrage objective. A common strategy is to reserve a portion of storage capacity for regulation during periods when price spreads are low, and dedicate the rest to arbitrage when spreads are high. The control algorithm must dynamically allocate capacity between the two uses based on real-time prices and regulation signals. This requires a more sophisticated optimization that includes the regulation revenue as a term in the objective function. Operators should be aware that regulation participation may increase wear and tear on chillers due to more frequent cycling; a maintenance budget increase of 10–20% should be factored in.

Long-Term Contracts and Hedging

To stabilize revenue and justify capital investment, operators may enter into long-term contracts with utilities or load-serving entities for load shifting capacity. These contracts typically pay a fixed monthly capacity payment plus a variable payment for energy shifted. They reduce exposure to price volatility but also cap upside. For risk-averse investors, a portfolio approach that combines long-term contracts with some speculative arbitrage capacity can provide a balanced risk-return profile. We recommend that at least 50% of storage capacity be contracted for the first few years of operation to ensure debt service coverage, with the remaining capacity used for opportunistic trading.

Risks, Pitfalls, and Mitigations: What Can Go Wrong and How to Avoid It

Like any trading strategy, latent load arbitrage carries risks. Some are technical, such as equipment failure or control system bugs; others are market-related, such as changes in tariff structures or price volatility. This section catalogs the most common pitfalls we have encountered and provides concrete mitigation strategies.

Pitfall 1: Over-Optimizing for Historical Price Patterns

Many operators tune their control algorithms to historical price data, only to find that the patterns shift when renewable penetration increases or market rules change. For example, in CAISO, the midday solar trough has deepened over time, but the timing of the evening peak has also shifted later as solar generation extends into the evening. A control algorithm trained on 2020 data may underperform in 2025. Mitigation: Use adaptive algorithms that continuously update their price forecasts based on recent data, and incorporate external inputs like solar generation forecasts and weather data. Re-train models at least quarterly.

Pitfall 2: Ignoring Thermal Comfort Constraints

In the pursuit of arbitrage revenue, operators may discharge storage too aggressively, causing indoor temperatures to drift outside comfort bands. This leads to tenant complaints and potential lease violations. Mitigation: Implement hard constraints on indoor temperature (e.g., ±2°F from setpoint) and incorporate a penalty term in the optimization for comfort violations. Use zone-level temperature sensors to detect hot spots early. In one case, a building manager had to pay $10,000 in compensation to tenants after a series of over-discharge events. A simple rule—never let storage state of charge fall below 15% during occupied hours—would have prevented the issue.

Pitfall 3: Underestimating Maintenance Costs

Thermal storage systems require regular maintenance: cleaning of heat exchangers, calibration of sensors, inspection of tank insulation, and replacement of pumps and valves. Ice storage systems, in particular, are prone to ice bridging and coil fouling. Operators often budget only 1–2% of capital cost annually for maintenance, but actual costs can be 3–5% for ice systems. Mitigation: Include a maintenance reserve fund in the project financial model. Conduct a preventive maintenance audit every six months. Consider a maintenance contract with the equipment vendor.

Pitfall 4: Market Rule Changes

Regulatory changes can undermine the arbitrage thesis. For example, a utility may introduce a demand charge that penalizes high peak demand even during off-peak hours, or a wholesale market may change its capacity product definitions. Mitigation: Stay informed through industry associations (e.g., ESA, NAESCO) and regulatory filings. Build flexibility into the control system to adapt to new rate structures. Diversify revenue streams across multiple markets (energy, capacity, ancillary services) to reduce dependence on any single product.

Pitfall 5: Sensor Drift and Data Quality Issues

Over time, temperature and flow sensors drift, leading to inaccurate state-of-charge estimates and suboptimal dispatch. Without reliable data, the optimization model becomes useless. Mitigation: Implement a sensor validation and calibration program. Use redundant sensors for critical measurements. Perform a monthly comparison of calculated state of charge against actual tank temperature profiles. If discrepancies exceed 5%, recalibrate immediately.

Decision Framework and Mini-FAQ: Key Questions for Practitioners

To help operators assess whether latent load arbitrage is right for their facility, we have compiled a decision checklist and answers to the most common questions we receive from practitioners.

Decision Checklist

• Price Spread: Does your facility face an on-peak/off-peak price differential of at least $50/MWh on average over the cooling season? If not, arbitrage may not be viable.
• Load Profile: Does your cooling load correlate with peak price periods? For example, commercial buildings with high afternoon loads in summer are ideal.
• Space Availability: Do you have adequate space for a storage tank? For chilled water, you need roughly 10 ft² per ton-hour of storage. Ice storage requires about 4 ft² per ton-hour.
• Capital Budget: Can you invest $200,000–$500,000 for a typical 1,000-ton-hour system? Financing options include energy service agreements (ESAs) and performance contracts.
• Technical Expertise: Do you have in-house staff capable of managing the control system and performing optimization? If not, consider a third-party operator.
• Market Access: Can you participate in wholesale markets directly or through an aggregator? If you are on a fixed retail tariff, you may need to negotiate a new rate structure.

Mini-FAQ

Q: Can I retrofit an existing chiller plant with thermal storage? Yes, in most cases. Retrofitting typically involves adding a storage tank and modifying the piping and controls. The existing chillers can be used for both direct cooling and charging. However, you must ensure that the chillers can operate at the lower temperatures required for ice storage (if applicable). A feasibility study is recommended.

Q: How long does it take to recoup the investment? Based on projects we have analyzed, simple payback periods range from 2 to 5 years for chilled water systems and 3 to 7 years for ice storage, depending on price spreads and utilization. With the addition of ancillary service revenue, payback can shorten by 6–12 months.

Q: What happens if the price spread disappears? If spreads narrow permanently (e.g., due to market design changes), the storage asset may become uneconomical. To hedge this risk, we recommend designing the system to also provide demand charge reduction or emergency backup cooling, so that it retains value even without arbitrage. Many operators use storage primarily for demand charge management and treat arbitrage as a secondary revenue stream.

Q: Do I need to be in a deregulated market? Not necessarily. Even in regulated markets, many utilities offer time-of-use rates or demand response programs that reward load shifting. However, the arbitrage potential is typically lower than in deregulated markets. A thorough analysis of your specific tariff is essential.

Q: How do I ensure my control system is secure from cyber threats? Use industry-standard cybersecurity practices: segment the storage control system from the corporate network, use encrypted communications, and regularly update firmware. Consider a third-party security audit before commissioning.

Synthesis and Next Actions: Turning Insights into Implementation

Latent load arbitrage represents a significant, underutilized opportunity for organizations that own or operate large cooling systems. By treating thermal storage as a trading asset rather than a static infrastructure component, operators can generate substantial revenue while contributing to grid stability and renewable integration. This guide has provided a comprehensive framework covering the strategic rationale, technical design, operational workflows, technology choices, scaling strategies, and risk management. The next step is to translate this knowledge into action.

Immediate Next Actions for Practitioners

First, conduct a preliminary screening of your facility using the decision checklist above. Gather 12 months of hourly electricity consumption data and corresponding price data (from your utility bill or market feed). Calculate the potential arbitrage revenue using a simple spreadsheet model that assumes storage can shift 50% of peak cooling load for 4 hours per day. If the annual revenue exceeds 10% of the estimated capital cost, proceed to a detailed feasibility study.

Second, engage a qualified engineering firm with experience in thermal storage and energy markets. Ask for references from projects of similar scale and complexity. The feasibility study should include a preliminary design, cost estimate, and pro forma financial analysis. Ensure the study considers multiple storage technologies and control strategies.

Third, if the feasibility study is positive, develop a detailed implementation plan. This should include procurement specifications, a project schedule, and a commissioning plan. Secure financing through internal capital budgets or third-party energy service agreements. Establish key performance indicators (KPIs) and a monitoring plan before the system goes live.

Fourth, once operational, continuously optimize. Use the first year of data to refine your control algorithms and validate your financial model. Explore opportunities to participate in ancillary service markets or aggregate with neighboring facilities. Regularly review market conditions and adjust your strategy accordingly.

This is not a set-and-forget investment. It requires ongoing attention, but the rewards—both financial and environmental—are substantial. We encourage readers to share their experiences and lessons learned with the broader community to advance the practice of latent load arbitrage.